1. INTRODUCTION

Combining time-resolved spectroscopy with optical microscopy opens up fascinating insights into the excited state dynamics of nanoscale materials and even individual nano objects [1–4]. High spatial resolution, on the other hand, implies tiny detection volumes reaching down to femtoliters making great demands on the detection sensitivity. Time-correlated single photon counting (TCSPC) that detects the sample’s photoluminescence (PL) stands out as a background-free and sensitive technique. Whereas TCSPC readily provides single molecule sensitivity, it typically suffers from limited temporal resolution given by the employed detector and electronics [5].

Pump-probe techniques, on the other hand, reach into the sub-10 fs regime by detecting pump-induced transient changes in the sample response [6]. Because they rely on the detection of small variations in the probe pulse intensity they require particularly stable pulsed lasers. Here, fiber lasers may provide an opportunity as compact ultrafast sources. For spectroscopy, facile wide-range tunability of the source is essential. In general, frequency conversion of the laser fundamental can be achieved using additional external instruments, such as optical parametric oscillators. In the case of fiber-based sources tunability can also be achieved utilizing the soliton self-frequency shift (SSFS) effect. Applications of SSFS-based fiber laser systems have been shown for a variety of spectroscopic techniques including CARS [7], three-photon excitation [8], and fluorescence/phosphorescence lifetime imaging [9] and have been proposed as potential turnkey systems for endoscopic imaging of tissue [10].

Interferometric scattering (iSCAT) microscopy exploits the interference between the electric field scattered by a nanoscale object and a reference electric field, typically the field reflected at the interface formed by the sample substrate and the medium the object is embedded in to generate a strong optical contrast. In this way, iSCAT achieves high detection sensitivities reaching down to single nanoparticles [11–14]. Combined with a pump-probe scheme transient iSCAT (TiSCAT) can provide ultrafast time-resolved data of single nanoobjects [4].

We have developed a new dual color fiber-based laser system for transient interferometric scattering microscopy and spectroscopy. Combining all necessary optical components for pulse manipulation in one single box leading to exceptionally stable beam pointing needed for high resolution optical microscopy of single nanoobjects. The system includes a pump arm emitting at 780 nm, which can be modulated by an internal acousto optic modulator (AOM) and which is prechirped to compensate the dispersion of the utilized microscope objective. The probe arm is continuously tunable between 810 and 960 nm using SSFS. Coupled to a scanning confocal optical microscope with high numerical aperture objective, the setup provides TiSCAT data with sub-300 nm spatial and 300 fs temporal resolution. After introducing the laser system and the microscope setup, we demonstrate time-resolved imaging and spectroscopy on monolayer ${{\rm MoSe}_2}$ as well as on single semiconducting (6,4) single-walled carbon nanotubes (SWCNTs) with nearly shot-noise limited sensitivity down to $3.2 \cdot {10^{- 6}}$ at low excitation and probe fluences. ${{\rm MoSe}_2}$ and SWCNTs were chosen as model systems because in both cases the exciton dynamics is known to occur on the sub-ps to ps time scale making these materials well suited for testing the performance of the developed laser system [15–19]. Moreover, the nanometer feature sizes of materials provide a means for testing the achieved spatial resolution.

2. EXPERIMENTAL SETUP

A. Laser System

In the following, the principle of the tunable two arm fiber laser system is illustrated. A sketch of the system is depicted in Fig. 1. The starting point is a fiber oscillator operating at a center wavelength of 1560 nm emitting pulses at a repetition rate of 76 MHz. The output is then split into two arms, which act as pump and probe source for the TiSCAT experiment described below. Both arms are amplified to an average power of several watts by a two-stage chirped pulse amplification system and temporally compressed afterward using grating compressors. A mechanical delay line located in the pump arm allows for adjusting the temporal delay of both pulse trains up to a maximal delay of 500 ps, making time-resolved experiments possible.

 

Fig. 1. Schematic of the fiber laser based system. (SHG, second-harmonic generation; TSHG, tunable second-harmonic generation; AOM, acousto optic modulator).

Download Full Size | PDF

After the amplification the pulses of the probe arm are coupled into a fiber (labelled as “shifting fiber” in Fig. 1) in which solitonic propagation due to a balancing between anomalous chromatic dispersion and self-phase modulation leads to temporally stable, transform-limited pulses. Due to the high peak intensities reached in the single-mode fiber and the broad spectral bandwidth of the short pulses, intra-pulse stimulated Raman scattering of the short wavelength part of the pulse spectrum takes place transferring energy to the longer wavelengths. This effect called “soliton self-frequency shift” (SSFS) was first described experimentally by Mitschke and Mollenauer in 1986 [20] and leads to a progressive propagation of the pulse spectrum to higher wavelengths whose speed (in the moving frame of a single pulse) is determined by the peak intensity for a given optical fiber. This gives rise to the tunability of the frequency-shifted pulses by adjusting the energy of the pulses coupled into the fiber.

In a last step, the probe pulses are frequency-doubled with high efficiencies up to over 50% using a second-harmonic generation unit with tunable phase matching (tunable second-harmonic generation, TSHG). In this configuration the accessible tuning range of the probe arm ranges from 810 nm up to 960 nm with output powers up to 280 mW, depending on the chosen center wavelength. The pulses exhibit a ${{\rm sech}^2}$-shape, which is typical for solitonic propagation and are nearly transform-limited with a pulse width of 75–100 fs (full-width at half maximum [FWHM]) and a bandwidth of around 13 nm (FWHM). Exemplary spectra and a typical interferometric autocorrelation of the tunable probe pulses are shown in Fig. 2.

 

Fig. 2. (a) Exemplary spectra of the continuously tunable probe arm spanning the range from 810 nm up to 960 nm. (b) Exemplary interferometric autocorrelation of the TSHG. From the autocorrelation a temporal pulse width of ${\sim}90 \; {\rm fs}$ (FWHM) was calculated.

Download Full Size | PDF

The pulses of the pump arm are amplified and compressed like in the probe arm and then frequency-doubled to a center frequency of 780 nm. For a better temporal resolution of the experiment, the pulses are prechipped to account for chromatic dispersion mainly introduced by the subsequent AOM and the high numerical aperture (NA) microscope objective (${\sim}4000\;{{\rm fs}^2}$). This leads to the shortest pulse width of the pump pulse at the position of the sample and thus to the best timing resolution of the experiment. The AOM can be modulated with frequencies up to 1 MHz for lock-in detection schemes and is used to continuously adjust the output power of the pump arm, which is up to ${\gt}{1}\;{\rm W}$. A pump-pulse modulation frequency of 100 kHz was found to yield the best balance between detector bandwidth and high sensitivity of the TiSCAT measurements.

The optical pathway of the laser system is predominantly guided in polarization-maintaining single mode fibers, ensuring persistent optical alignment and pulse characteristics. The whole laser system is housed in a single box containing the fiber amplifiers, the motorized delay line, and the AOM as well as the temperature stabilized optical pathways of the pulse compressor, frequency conversion, and fiber coupling unit. This improves the beam pointing stability of the system, which is crucial for experiments that rely on the spatial overlap of two beams. External noise source influences are minimized in this way as well. The relative intensity noise (RIN) of the probe arm was measured to be 0.14% when integrated from 10 Hz to 10 MHz with a value of around ${-}{130}\;{\rm dBc/Hz}$ at a frequency of 100 kHz, which is the modulation frequency of our TiSCAT setup (Supplement 1 Fig. S3). This represents a significant noise reduction as compared with our previously used system based on a Ti:Sa laser-pumped photonic crystal fiber for probe light generation in [4] with ${\rm RIN} = - {85}\;{\rm dBc/Hz}$ (Supplement 1 Figs. S2 and S3). Directly after leaving the laser housing, both arms are spatially overlapped, expanded, and sent to the microscope without any need for further beam manipulation to maintain high pointing stability.

B. TiSCAT Microscopy

The experimental setup for confocal TiSCAT and PL microscopy is depicted in Fig. 3. The spatially overlapped and expanded probe and pump beams are focused onto the sample by a high-NA (${\rm NA} = {1.49}$) oil immersion objective in a commercial inverted scanning microscope. The diffraction-limited focal width (FWHM) in the spectral range of 800 nm used in the experiments is thus $d = 0.51\lambda /{\rm NA} = 274\;{\rm nm} $. The reflected and backscattered light is collected after transmission through a 30/70 beam splitter and spectral filtering of the residual pump-light. The intensity of the probe light is then measured by a fast silicon photodetector with a bandwidth of 400 kHz. The pump-induced change of the detected intensity is measured by a lock-in detection scheme in which the signal is demodulated at the AOM frequency. In general, to maximize the signal-to-noise ratio, the modulation frequency should be chosen as high as possible to avoid low-frequency perturbations and minimize the 1/f-noise of a laser source. Typically, MHz modulation frequencies are applied for this reason [21]. In our case, on the other hand, we chose 100 kHz as a compromise between a low noise measurement and a slower but more sensitive photodetector. With this choice, we are able to detect transient signal variations with probe fluences as low as $8 \cdot {10^4}$ photons per pulse arriving at the sample with high sensitivity down to $3.2 \cdot {10^{- 6}}$, as we show below. We note that higher sensitivities could be achieved by measuring in transmission mode because the detected probe intensity would be increased by a factor of around 30 due to the higher transmittance and because the beam splitter could be omitted in this configuration. This increase in probe intensity would translate into a reduction of the root-mean square (rms) shot-noise level by factor 5.5, as it scales with the square root of the detected power [2]. However, this configuration would have the drawback of a far more complex experimental setup and prevents the potential extension of the experiment by applying an additional near-field probe for nanoscale resolution measurements (e.g., [22]).

 

Fig. 3. Experimental setup for transient interferometric scattering (TiSCAT) and PL microscopy. (APD, avalanche photodiode; PD, photodiode).

Download Full Size | PDF

The instrument response function (IRF) of the TiSCAT experiment was measured by cross correlation of pump and probe arm on an iron iodide sample as it gives an instantaneous nonlinear sum frequency signal with high efficiency [23]. The transient is shown in Fig. 4 as well as a Gaussian fit (red line), which yields a width of 285 fs (FWHM). With this temporal resolution, decay dynamics well below a few hundred fs can be studied by fitting the convolution of the appropriate model function and the IRF curve to the measured data [24]. The detection scheme can be changed from the TiSCAT measurement on the photodiode (PD) to PL imaging with a single photon counting avalanche photodiode by flipping a mirror and disabling the probe arm emission.

 

Fig. 4. Instrument response function of the TiSCAT setup with a temporal width of 285 fs (FWHM) measured by sum frequency generation on iron iodide nanocrystals.

Download Full Size | PDF

For ultrathin layers on a transparent substrate producing a moderate contrast signal, the reflectance contrast is mostly determined by the sample absorption [25,26]. The signal detected in TiSCAT of semiconductor monolayers and single SWCNTs on glass can thus be primarily attributed to transient absorption [4]. By implementing a lock-in detection scheme the pump-induced changes in the absorption cross section $\sigma$ can hence be measured as

C. Sample Preparation

Single-walled carbon nanotubes were individualized by stirring the CNT raw powder in 1% sodium deoxycholate (DOC) solution for five weeks. Single (6,4) SWCNTs were then separated by aqueous two-phase separation, as described by Subbaiyan et al. [27]. Samples for microscopy were produced by spin-coating 30 µl of the solution on glass cover slides at 3000 rpm for 60 s. Monolayer molybdenum diselenide (${{\rm MoSe}_2}$) samples were obtained by mechanical exfoliation and polymer-assisted transfer onto a glass cover slide covered by 40 nm of hexagonal boron nitride (hBN), as described in [28]. After transfer, polymer residues, interfacial bubbles, and wrinkles were efficiently removed by vacuum-annealing at 200°C, resulting in a stacked layer formed by ${{\rm MoSe}_2}$ on hBN on glass without a gap (Supplement 1 Fig. S5).

 

Fig. 5. (a) PL image of ${{\rm MoSe}_2}$ detected for pulsed excitation at 780 nm. (b) TiSCAT image of the same sample area recorded at zero delay time between pump pulse at 780 nm and probe pulse at 810 nm. The monolayer in the center and the multilayers in the left part of the image exhibit a negative TiSCAT signal corresponding to pump-induced absorption. Several point-like objects can be observed in both areas with positive signals due to pump-induced ground state bleaching. The signal decay for monolayer ${{\rm MoSe}_2}$ is significantly faster than for multilayers, as can be seen from the pump-probe images at (c) 20 ps and (d) 100 ps delay, respectively.

Download Full Size | PDF

3. EXPERIMENTAL DATA AND DISCUSSION

A. Measurements on MoSe₂

The optical and electronic properties of transition metal dichalcogenides (TMDCs) depend on the number of stacked layers of the material [29]. ${{\rm MoSe}_2}$ transforms from an indirect semiconductor to a direct semiconductor, as it is reduced from multilayer to a monolayer sheet [30]. In Fig. 5, this can be seen by comparing the PL image (a) and the TiSCAT image at zero time-delay $t = {0}\;{\rm ps}$ (b) of the same sample region. In both measurements, the material was excited at 780 nm using $2.6 \cdot {10^5}$ photons per pulse for PL and $1.0 \cdot {10^6}$ photons per pulse for TiSCAT, respectively. In the PL image, only the monolayer can be observed, as the multilayers do not show radiative recombination and thus remain dark. In this way, the monolayer ${{\rm MoSe}_2}$ can be clearly identified. In contrast, the TiSCAT image taken at a probe wavelength of 810 nm with $1.3 \cdot {10^5}$ photons per pulse exhibits strong negative signals for both mono- and multilayer regions. The acquisition time of the image shown in Fig. 5(b) was ${256} \times {256} \times {0.016}\;{\rm s} = {1049}\;{\rm s}$.

The monolayer part in the middle of the image shows similar features in the TiSCAT and in the PL image, which can be assigned to wrinkles, cracks, or regions with defects or doping. The monolayer TiSCAT signal taken for a probe wavelength of 810 nm is negative at zero delay corresponding to an increase of absorption due to the pump excitation. This photoinduced absorption (PA) signal is in agreement with literature results on TMDCs showing a transient increase in absorption on the longer wavelength side of the A-excitonic peak, which is centered at 785 nm for ${{\rm MoSe}_2}$ [26,31]. This PA is attributed to a shift of the exciton resonance due to excitation-induced bandgap renormalization and broadening of the exciton peak caused by excitation-induced dephasing [25,32]. Additionally, in Fig. 5, small spots with positive signals are visible revealing a pump-induced decrease of ground-state absorption. This could potentially be attributed to local defect sites or doping, causing a spectral redshift of the exciton resonance and leading to a ground-state bleaching contribution at 810 nm, as discussed in more detail below.

TiSCAT images taken at increasing delay times of $t = {20}$ and 100 ps, respectively (Fig. 5), reveal different decay dynamics for mono- and multilayer regions and within these regions. Whereas the signal in the monolayer decays within 100 ps, the signal in the multilayer regions persists for longer delay times, thus reflecting the layer-dependent bandgap characteristics. Remarkably, several features in the monolayer region can also be observed for longer times, highlighting the importance of spatially resolved probing for a comprehensive sample characterization.

Figure 5 illustrates the improvement of the spatial resolution of the TiSCAT measurement in contrast with the PL image. The pump-probe technique exhibits the nonlinear characteristics of a multiphoton measurement; thus, the effective point spread function (PSF) is created by the multiplication of the pump PSF and the probe PSF [33]. For the present measurement with excitation at 780 nm and probing at 810 nm, this yields a diffraction-limited spatial resolution of 200 nm, about a factor $\sqrt 2$ smaller than for linear microscopy such as one-photon excitation photoluminescence. The experimentally observed FWHM of 234 nm shown in Fig. 7(c) for a slightly larger probing wavelength of 880 nm confirms this discussion.

To learn more about the excited dynamics and the origin of the transient signal TiSCAT, measurements were carried out for different probe wavelengths in a broad spectral range, reaching from 810 to 940 nm at a fixed sample position in the middle of the monolayer flake. For all probing wavelengths, an instantaneous signal rise within the IRF response is observed. Upon close inspection, the transients show an additional oscillatory component with a period of about 300 fs. Figure 6(d) presents the TiSCAT signal detected at 810 nm within the first 2 ps together with the fitted model function after convolution with the measured IRF. The oscillatory component can be assigned to the excitation of a strongly damped coherent phonon mode [34,35]. All transients, besides the one at 810 nm, can be described by a bi-exponential decay function with decay times of 3 and 22 ps, respectively [Figs. 6(b) and 6(c)]. Following the discussion in [25], the two decay times can be assigned to electronic and phononic relaxation upon pulsed excitation, respectively.

 

Fig. 6. (a) TiSCAT transients detected on monolayer ${{\rm MoSe}_2}$ at the position marked by “x” in Fig. 5(b) for different probe wavelengths between 810 and 940 nm. The black dotted lines mark the time delay where the images in Fig. 5 were taken. (b) Same transients as in (a) normalized to the same amplitude (${-}1$) for comparison. (c) Transient detected at 940 nm together with fitted model function (see text). (d) Transient detected at 810 nm together with fitted model function, including a damped oscillatory component for the initial dynamics.

Download Full Size | PDF

At a probe wavelength of 810 nm, a transition from pump-induced absorption to photobleaching occurs at a pulse delay of around 50 ps. An explanation for this behavior could be an overlap of PA and photoinduced bleaching (PB) signal contributions with different excited state dynamics. As mentioned, the excitation wavelength of the A-exciton is located at around 785 nm and was reported to show transient ground-state bleaching due to Pauli-blocking [36]. The small PB signal observed at 810 nm probing for longer delay times can thus be attributed to a small residual fraction of longer-lived excited states preventing complete ground-state recovery. These states could be trions formed by environmental doping, e.g., by polymer residues from flake transfer [37]. Larger signatures of such doping and the resulting slow decay can be seen in Fig. 5(c), 100 ps after excitation as more localized positive signals.

B. Measurements on Individual (6,4) SWCNTs

Individualized (6,4) single-walled carbon nanotubes on glass were identified by photoluminescence microscopy with pulsed excitation at 780 nm and an excitation power of around 1.4 µW corresponding to $7.2 \cdot {10^4}$ photons per pulse. Their (6,4) chirality was confirmed by analysis of the emitted PL spectrum which is centered at around 900 nm [see Fig. 8(b), orange curve] [27]. An additional emission peak located at 970 nm is assigned to defect states [38]. These defects could result from the separation process of the carbon nanotubes or could be induced by irradiation with the excitation pulses, as reported for this type of nanotube [38]. The second point would again emphasize the importance of minimizing laser power. In the course of the presented measurements, no significant bleaching of the PL intensity was observed.

 

Fig. 7. (a) Confocal PL image of a single (6,4) SWCNT on glass detected at 880 nm upon pulsed excitation at 780 nm. (b) TiSCAT image at a pump wavelength of 780 nm and a probe wavelength of 880 nm at zero time delay. (c) Cross section of the TiSCAT image shown in (b). The fit of a Gaussian function (red line) to the cross section exhibits a FWHM of 234 nm.

Download Full Size | PDF

 

Fig. 8. (a) TiSCAT transient detected at a probe wavelength of 880 nm of the single (6,4)-SWCNT shown in Fig. 7. Excitation and probing were done with $1.4 \cdot {10^5}$ photons/pulse and $7.6 \cdot {10^4}$ photons/pulse, respectively. (b) Photoluminescence spectrum of the (6, 4)-SWCNT shown in Fig. 7 (orange curve) and measured peak transient absorption signal for different probe wavelengths depicted as black circles. A Voigt function (blue line) is fitted to the TiSCAT data taking the Gaussian laser spectrum into account.

Download Full Size | PDF

Figure 7(b) presents the first TiSCAT image of a single (6,4) SWCNT. For the measurement, the pump and probe power at the sample were adjusted to approximately 1.4 µW equivalent to $1.4 \cdot {10^5}$ photons/pulse (pump, modulated with 50% duty cycle) and $7.7 \cdot {10^4}$ photons/pulse (probe). Both laser beams were polarized parallel to the nanotube axis for highest PL intensity and TiSCAT contrast. Again, the same features are present in both images. The spatial resolution of the TiSCAT experiment can be determined from cross sections taken perpendicular to the nanotube axis. The width of the cross section shown in Fig. 7(c) indicates a spatial resolution of 234 nm, slightly higher than the theoretical prediction assuming optimum focusing conditions. Remarkably, the nanotube cannot be detected in the simultaneously recorded elastic scattering image due to the dominating background intensity contribution resulting from the reflection at the air–glass interface (Supplement 1 Fig. S1).

The transient depicted in Fig. 8(a) was taken in the middle of the SWCNT and shows a peak value consistent with that observed in the TiSCAT image in Fig. 7(b). The signal decay can be modeled by combining contributions from ultrafast exciton-exciton annihilation (EEA) and nonradiative exponential decay [4]. For the present transient, we determine a diffusional time of 140 fs, determining EEA and an exponential decay time of 3.4 ps. The exponential decay time is significantly shorter than the typical times observed for single (6,5) SWCNTs [4], which can be attributed to the presence of defect sites leading to quenching as noted [38].

For an estimation of the noise characteristics of the experiment, the signal for negative times (the probe pulse arriving at the sample before the pump pulse) was used. The rms-noise was calculated as the standard deviation of the signal to be ${\sigma _{\rm{exp}}} = 3.2 \cdot {10^{- 6}}$. The theoretical rms shot-noise ${\sigma _{\rm{shot}}}$, which represents the limit in detection sensitivity, was calculated as ${\sigma _{\rm{shot}}} = \sqrt {2h \nu BP} = 0.10\;{\rm pW}$ [2]. Here, we took typical experimental parameters used for dynamic measurements, which are a probe power of ${P} = 1.4\;{\unicode{x00B5}}{\rm W}$ (of which 5% reach the detector), a probe wavelength of 880 nm, and a measurement bandwidth of 1/3 Hz. This translates into a detection sensitivity limit of ${\sigma _{\rm{shot}}}/P = 1.4 \cdot {10^{- 6}}$. Thus, up to a factor of ${\sim}2.3$, the measurement is approximately shot-noise limited. The additional noise sources could be explained by noise of the detection electronics and additional 1/f-noise of the laser source still present at this modulation frequency. As the rms shot-noise level scales with the square root of the incident probe power while the signal is linearly dependent, the detection sensitivity could be improved with higher power levels (as shown for example in [2]), but this also carries the risk of photodamage to the sample. Taking the rms noise level of the experiment into account, the signal-to-noise ratio at the delay time with the highest signal was calculated to be 23.

TiSCAT spectra of single (6,4) carbon nanotubes were measured by step-by-step wavelength tuning of the center wavelength of the probe arm, as shown in Fig. 8(b) (black data points). The TiSCAT spectrum was extracted from the maximum amplitude of the corresponding TiSCAT signal at each probe wavelength. Additionally, the PL spectrum is shown in the figure as an orange data curve. The TiSCAT spectrum is fitted with a Voigt function. This model function is chosen because the measured signal results from a convolution of the tuned laser spectra, which have a bandwidth of approximately 13 nm, and the absorption spectrum of the investigated lowest exciton state ${S_{11}}$, which is assumed to exhibit a Lorentzian line shape [39]. The peak position of 1.4 eV matches the value reported for the ${S_{11}}$ absorption of (6,4) nanotubes in the literature [40]. We thus attribute the TiSCAT spectrum of individual (6,4) SWCNTs to bleaching of the ground-state absorption. The Stokes shift between ground-state bleaching and photoluminescence peaks of 30 meV is in agreement with literature reports [41] as well.

4. CONCLUSION

In summary, we have developed a new wavelength-tunable femtosecond fiber-based laser system with two synchronized outputs suitable for the investigation of the photoinduced dynamics in single nanoscopic objects. Using the two outputs as pump and probe beam, respectively, and coupling them to a scanning confocal microscope subpicosecond time-resolved images and transients can be obtained with 300 nm spatial resolution. We have demonstrated the performance of the measurement platform using single-layer ${{\rm MoSe}_2}$ and single semiconducting (6,4) carbon nanotubes as model samples. TiSCAT measurements of ${{\rm MoSe}_2}$ flakes provided snap shots of the layer-dependent and spatially heterogeneous decay dynamics at distinct points in time after pulse laser excitation. Wavelength-resolved transients of monolayer ${{\rm MoSe}_2}$ showed spectrally uniform decay of the transient absorption signal related to the A-exciton over a broad spectral range of 130 nm, with the exception of an additional longer-lived ground-state bleaching contribution close to the A-exciton resonance. TiSCAT measurements on single (6,4) single-walled carbon nanotubes on glass revealed the ultrafast exciton decay in this type of nanotubes for the first time. Here, the tunability of the probe arm was used to detect the first transient absorption spectrum of a single semiconducting SWCNT. The detection sensitivity of the TiSCAT system was shown to be close to the shot-noise limit, enabling time-resolved measurements at low excitation powers. Due to the high output power of the fiber laser-based system, its frequency range could be further extended by additional nonlinear frequency conversion, such as SHG. This would significantly increase the range of accessible materials. Further, with the included pump-pulse intensity modulation, the tunable laser system could be utilized also for stimulated Raman scattering spectroscopy and microscopy.